You are on page 1of 14

Nuclear Engineering and Design 192 (1999) 271 284

www.elsevier.com/locate/nucengdes

Seismic response of buried pipelines: a state-of-the-art


review
T.K. Datta *
Department of Ci6il Engineering, Indian Institute of Technology, Hauz Khas, New Delhi 110016, India

Abstract
A state-of-the-art review of the seismic response of buried pipelines is presented. The review includes modeling of
soilpipe system and seismic excitation, methods of response analysis of buried pipelines, seismic behavior of buried
pipelines under different parametric variations, seismic stresses at the bends and intersections of network of pipelines.
pipe damage in earthquakes and seismic risk analysis of buried pipelines. Based on the review, the future scope of
work on the subject is outlined. 1999 Elsevier Science S.A. All rights reserved.

1. Introduction
The seismic behavior of buried pipelines and
piping systems is quite different from that of the
above-ground structures in many respects, such
as:
The horizontal inertia forces, which comprise
the main factor affecting the design of aboveground structures, are largely resisted by the
surrounding soil in the case of buried pipelines.
For above-ground structures, the foundation is
usually assumed to follow the ground motion
and, therefore, the relevant response is displacement relative to the foundation. In the
case of buried pipelines, the relative movement
between the pipe and the surrounding soil is
responsible for inducing stresses at the joints.

* Tel.: +91-11-686-7754; fax: + 91-11-686-2037.


E-mail address: tkdatta@netearth.iitd.ernet.in (T.K. Datta)

The ground motion is considered to be coherent for most over-ground structures, while for
buried pipelines it is considered as incoherent
because of the phase difference between different stations and the change in shape due to the
variation of soil properties along the pipeline.
The damage of one over-ground structure is
generally restricted to that structure alone, but
the damage at a certain location within a network of pipelines will affect other portions of
the system.
Field observations and various studies indicate
that major seismic hazards to buried pipeline systems are: (1) excessive axial and bending stresses
and deformations in pipelines created mainly by
the phase difference and change of wave shape
between different points along the pipeline; (2)
large displacements resulting from the fault movement during an earthquake if the pipeline crosses
a major fault; and (3) landslides and buoyancy
caused by soil liquefaction.

0029-5493/99/$ - see front matter 1999 Elsevier Science S.A. All rights reserved.
PII: S 0 0 2 9 - 5 4 9 3 ( 9 9 ) 0 0 1 1 3 - 2

272

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

Damage and disruption of buried pipelines


caused by an earthquake may have severe effects
on civil life since it may lead to loss of vital
services, communications and transportation systems. As a consequence, the seismic behavior of
these structures has been investigated by many
researchers. A number of issues are involved in
the study of seismic analysis and behavior of
buried pipelines. They include: (1) modeling of
seismic excitation and soil pipe system; (2) evaluation of seismic stresses in the pipelines; (3)
stresses at the junctions and bends; (4) pipeline
stresses due to fault movement and soil liquefaction; (5) pipeline damage in earthquakes; and (6)
vulnerability of pipelines to seismic hazards. In
this paper, a state-of-the-art review on seismic
response and behavior of buried pipelines, addressing the above issues, is presented.

2. Modeling of soil pipe system and seismic


excitation
With the postulation that the soil does not lose
its integrity during the design earthquake, the
basic concept governing the response of underground pipelines is that the soil is stiff compared
to the structure. Therefore, the earthquake deformation is imposed on the structure, which must
conform to this deformation. These deformations
are of two types curvature and shearing. The
former represents the direct imposition of the soil
curvature on the structure, which must have the
capacity to absorb the resulting strains. The latter
represents the lag of soil in response to the base
acceleration imparted to it through the bedrock.
It is important for the designer to recognize that
the effect of the earthquake on underground
pipelines is the imposition of an arbitrary deformation which can not be changed by strengthening the structure (so long as the soil is stiffer than
the structure). The structural design criterion is,
therefore, a provision of sufficient ductility to
absorb the imposed deformation without losing
the capacity to carry static loads, rather than a
criterion for resisting inertial loads at a specified
stress level.

It should also be recognized that the absolute


amplitude of the earthquake displacement may be
large, but this displacement is spread over a long
length. Therefore, the ratio of the earthquake
distortion is generally small, and usually within
the elastic deformation capacity of the structure.
If it can be established that the maximum deformation imposed by the earthquake will not strain
the structure beyond the elastic range, no further
provision to resist the deformation is required. If
certain joints are strained into the plastic range,
the ductility of such joints should be investigated.
Soilstructure interaction becomes important
when the soil stiffness is comparable to the pipelines stiffness. Under such a condition, a rigorous
dynamic analysis for the soilstructure system is
required. Seismic response analysis of buried
pipelines is a complex phenomenon including
three-dimensional dynamic analysis of the soil
structure system under multipoint seismic excitation and treatment of non-homogeneous
non-linear/linear viscoelastic medium. A rigorous
analysis satisfying all these conditions is almost
impossible; therefore, different degrees of simplifications are made to obtain a good estimate of the
response quantities of interest. The level of simplification depends upon the response quantities
of interest, and it primarily governs the type of
modeling of the system. Different types of modeling of the underground pipelines are available,
starting from extremely simple to complex threedimensional modeling of the soilstructure system. Accordingly, the method of analysis varies.
The simplest model of buried pipelines is the
one in which no soilstructure interaction is accounted for. The structure is assumed to follow
the deformation of the ground. When the soil
structure interaction is considered, models that
have been widely used are shown in Figs. 14.
Fig. 1 shows the model of a beam on elastic
springs (they may be also in axial direction) to
represent long buried pipelines in which bending
and axial deformations are of main concern. A
shell model for a large diameter buried pipeline is
shown in Fig. 2. The shell model is assumed to be
resting within a viscoelastic medium. In Fig. 3, the
cross-section of a large diameter buried pipeline is
shown. This is a plane-strain model of the struc-

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

273

Fig. 3. Plane-strain model.

Fig. 1. Beam model in elastic foundation.

ture. Fig. 4 shows a hybrid model for a cylindrical


tunnel. This type of model is applicable for large
diameter pipelines.
In Fig. 1, the spring represents the soil stiffness.
The soil damping and stiffness are calculated separately and are included in the mathematical
model by equivalent springs and dampers. The
shell model shown in Fig. 2 provides both peripheral and longitudinal stresses in the structure due
to earthquake waves incident at an angle with the
longitudinal axis of the structure as well as at an
angle with one of the principal axes of the shell
cross-section. The shell model is also capable of
predicting the buckling failure of the structure.
The plane-strain model, as shown in Fig. 3, is
widely used whenever the hoop stress and the
radial displacement are to be obtained due to

Fig. 2. Shell model.

seismic waves. The possibility of the cross-section


buckling is also investigated using this model. In
the hybrid model shown in Fig. 4, the interior
region (R1) is modeled by the finite element
method (FEM), while the outer region (R2) is
modeled by the half space continuum. A planestrain model is adopted for both regions. Continuity of displacement and strain is maintained at
the interface boundaries between the two regions.
Many analyses of buried pipelines have been
done with P, S or Rayleigh waves with ground
displacement modeled as harmonic functions.
Waves propagated parallel to the long axis of the
pipelines tend to enforce a corresponding sinusoidal transverse distortion on the structure for
S-waves. S-waves traveling at right angles to the
structure tend to move it back and forth longitudinally, and may tend to pull it loose at zones of

Fig. 4. Hybrid model.

274

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

abrupt transition in soil conditions, where wave


properties may vary. Diagonally impinging waves
subject different parts of the pipeline to out-ofphase displacements. This results in a longitudinal
compression-rarefaction wave that travels along
the structure. Seismic shear waves may have any
inclination, but observations indicate that waves
causing displacements in a horizontal plane usually have the greatest amplitude; the amplitude of
vertical waves is generally one-half to two-thirds
as great as that of the corresponding horizontal
waves. The velocity of propagation of shear waves
decreases with decreased rigidity of the rock or
soil. Correspondingly, the amplitude of the vibration increases as the medium becomes softer.
For buried pipelines, the ground motion becomes incoherent because of the phase difference
between two points and change of wave shape
along the structure. Therefore, the single response
spectrum is not applicable for such long systems.
For such cases, the time history of ground acceleration/displacement is useful for the deterministic
seismic response analysis with assumed direction
and velocity of wave propagation. For non-deterministic analysis, the randomness of ground motion is mostly characterized by a power spectral
density function (PSDF) of ground displacement/
acceleration. Clough and Penzien (1975) and others considered the strong ground motion as a
white-noise process of limited duration. However,
a stationary filtered white-noise of limited duration is found to be more representative of actual
ground motion. The phase difference between
ground motions at two points is considered by
defining a correlation function involving the separation distance between the two points the shear
wave velocity and the frequency of ground motion. For non-stationary random ground motion,
it is generally modeled as a uniformly modulated
non-stationary random process. Various types of
modulating functions are used to model different
types of non-stationarity.

of analysis have been presented by researchers to


obtain the seismic response of buried pipelines.
Methods of analysis existing in the literature vary
from simplified analysis to complex three-dimensional finite element methods. Available methods
of analysis can be grouped under the categories
discussed in the following sections.

3.1. Analysis without considering soilstructure


interaction
Deformations in long structures are created
mainly due to the phase difference (time lag) and
the change of wave shape between different stations along the structure. Newmark and Rosenblueth (1971) presented simplified methods for
calculating the pipe deformations under the assumptions that (1) inertia forces and soilpipe
interaction are neglected (i.e. the pipe and soil
move together exactly), and (2) earthquake excitation may be modeled as a travelling wave having
constant wave shape (i.e. the time history of two
stations along the propagation path differs only
by time lag). Using these assumptions, they obtained the free field strain and curvature. In the
case of a general angle of incidence of the earthquake wave, Hindy and Novak (1979) proposed
other expressions for free field strain and curvature due to two types of waves, namely P-waves
and S-waves. Kuesel (1969) applied this method
to the seismic design of the San Francisco
subway.
Shah and Chu (1974) developed a simplified
approach for determining seismic stresses. The
simplified modeling and the formulae derived are
based on Newmarks values of seismic strain for
buried pipelines. The formulae include bending
moment and shear force for (1) straight elements
(2) bends and (3) T-junctions. The approximate
formulae are not valid for the ratio of wavelength
to length of the element less than 3p/4.

3.2. Quasi static analysis with soilstructure


interaction
3. Methods of analysis
Depending upon different types of model and
response quantities of interest, different methods

In this analysis, the soil deformations are


matched by a combination of pipe deformations
and relative deformations of joints. Using this

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

approach, upper bounds of axial pipe strain and


pipe curvature at a particular point are obtained
by Newmark and Rosenblueth (1971) and Hall
and Newmark (1977) by neglecting the relative
deformations at the joints (i.e. the pipe is assumed
to be flexible with respect to the soil). Upper
bounds of the relative displacements and relative
rotations at the joints are given by Wang et al.
(1979) by ignoring the pipe segment deformations
(i.e. the pipe is rigid with respect to the soil).
Assuming that the soil displacement in the axial
direction is matched only by the joint displacement and no slippage occurs between the pipe and
the soil, Wang et al. (1979) and Nelson and
Weidlinger (1979) presented two analyses for the
axial displacement of long segmented buried
pipelines due to traveling waves. The analysis
used the approach of a beam on an elastic foundation (Fig. 1), and the straight pipeline is modeled by a set of rigid segments connected by
flexible joints represented by elastic springs and
dashpots. In another study, Wang et al. (1982)
used a model in which pipe segments have finite
stiffness in the axial direction (semi-rigid). Consequently, the soil strain is matched by the combination of pipe strain and the relative displacement
of joints. They derived expressions for the pipe
strain ratio and the relative joint displacement
ratio. Nelson and Weidlinger (1979) developed an
interference response spectrum (IR spectrum), in
which any particular ordinate represents the absolute maximum axial out-of-phase response between two adjacent joints due to the incoherent
ground motion resulting from the phase delay.
The IR spectrum was constructed by neglecting
pipe joint stiffness and damping. Nelson and
Weidlinger (1979) also used the IR spectrum to
define the amplification factor for the displacement of joints of buried pipelines. The idealization of a buried pipeline as a beam on an elastic
foundation in axial and lateral direction was used
by Kubo et al. (1979) to realize one formula for
calculating the normal stresses, using the axial
stresses caused by the axial deformations and that
caused by bending deformations. Novak and
Hindy (1980) modified the equations given by
Hindy and Novak (1979) by introducing reduction factors accounting for the effect of soilpipe

275

interaction to obtain a quasi static analysis. Singhal and Zuroff (1990) proposed a quasi static
analysis, using the theory of beam on elastic
foundation, to obtain the response of buried
framed structure with flexible joints under earthquake excitation.

3.3. Dynamic analysis by considering the theory


of beam on elastic foundation
Considering the pipeline as a long continuous
(without joints) system, a spring supported
lumped mass model for such pipes was employed
by Hindy and Novak (1979). A deterministic
analysis of the response (both axial and lateral) of
buried pipelines to fully correlated traveling
waves, described by a single time history travelling along the pipe axis, was performed. The
soilpipe interaction was incorporated by a
spring dashpot system whose reactions were
derived from the static and dynamic continuum
theories. They also obtained the effect of pipe
joints on the response by modifying the axial
stiffness matrix of the pipe to incorporate the
axial stiffness of the joints. A similar analysis was
carried out by Yuan and Walker (1970) and
Wong and Weidlinger (1983). Using a spring supported beam model (Fig. 1), Hindy and Novak
(1980) obtained the response of buried pipeline
for random ground motion characterized by a
PSDF and a cross-spectral density function. The
soil reactions were derived from the static and
dynamic continuum theories and modal spectral
analysis was employed to evaluate the pipe response in both axial and lateral directions. Datta
and Mashaly (1986) extended the work of Hindy
and Novak (1980) to incorporate the effect of
cross terms of soil stiffness and damping matrices
in the stochastic seismic analysis of buried
pipelines.

3.4. Analysis by considering shell theory with


soilstructure interaction
Most of the work done on the seismic response
of buried pipelines used the beam theory. However, in such a theory the displacements are uncoupled and the beam model (Fig. 1) can not

276

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

describe the buckling and fracture phenomena.


These phenomena are best described by a shell
model which gives displacements in the three directions. In the response analysis of buried pipelines due to traveling waves, the pipe has been
modeled as a thin isotopic elastic cylindrical shell
in a viscoelastic medium. Muleski et al. (1979a,b)
used the Flugge bending theory of shell to develop
three decoupled equilibrium equations giving the
displacement in three directions due to axial
ground motion. Datta et al. (1981) and OLeary
and Datta (1985a) proposed axisymmetrical and
three-dimensional analysis (respectively) for the
dynamic response of buried pipelines to incident
compressional waves (S and P) traveling along the
pipeline, with low frequencies and long wavelengths. Wong et al. (1986) and Luco and Barnes
(1994) also carried out an analysis of pipeline by
shell theory.

3.5. Analysis using dynamic plane-strain model


with soil structure interaction
Wong et al. (1986) demonstrated the importance of considering the effect of pipeline motion
in modifying the free field ground motion into the
pipeline analysis. The pipeline modeled as continuous beams or cylindrical shells on an elastic
foundation do not consider this effect and as a
result the full soil structure interaction effect is
missing in such idealization. OLeary and Datta
(1985b), Datta et al. (1984), Wong et al. (1986)
and Takada and Tanabe (1987) studied this full
soilpipe interaction problem in three-dimensional response behavior of pipelines. The pipeline
was subjected to plane body and surface waves
moving at an arbitrary angle to the axis of the
pipeline in a semi-infinite medium. The equations
of elasto-dynamics governing the motion of the
pipe and surrounding ground were solved using
cylindrical eigen functions. The solution outside
the pipeline was expressed as the sum of a complete expansion in terms of outgoing waves satisfying the boundary conditions on the free ground
surface and the incident waves (Fig. 3). This
technique enabled the outer solution to be
matched with a finite element representation inside for the structure (if needed). An exact method

of solution involving series expansion of incident


and reflected SH waves of cylindrical wave function was presented bv Lee and Trifunac (1979) for
circular tunnels (or big diameter pipelines). It
used Hankel functions of the second kind in the
series expansion for the scattered and diffracted
waves (reflected wave function). The differential
equation of the total wave was solved under the
boundary conditions imposed for the whole
structure.

3.6. FEM analysis


Dynamic stresses and displacements of a cylindrical tunnel (or big diameter pipeline) embedded
in a semi-infinite elastic medium were analyzed
for P, SV and Rayleigh wave by Wong et al.
(1985) and Masso and Attalla (1984). The problem was considered as one of plane strain, in
which the wave propagates perpendicular to the
longitudinal axis of infinitely long tunnel. A numerical technique that combines the finite element
method with the eigen function expansion was
used. The eigen function expansion of the free
field displacement and the scattered wave displacement was used outside the boundary of the
circular region of the soiltunnel system. Inside
the boundary, a finite element representation of
the soilstructure system was adopted (Fig. 4).
The continuity of displacement and truncation
was imposed at the boundary.

3.7. Analysis of buried pipeline response to fault


mo6ement and soil liquefaction
The large abrupt differential ground motion on
an active fault represents one of the most severe
earthquake effects on a buried pipeline crossing
such a fault. The assessment of fault hazards
requires information about: (1) location of the
fault and the total width of the zone; (2) the
magnitude and direction of the expected displacements and their type (horizontal or vertical); (3)
geometry of the fault plane; and (4) the magnitude of the causative earthquake and its return
period. Newmark and Hall (1975) developed a
procedure for analyzing the effect of large fault
movements on buried pipelines. They related the

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

soil-slip friction on the pipe to the earth pressure


at rest (neglecting the passive soil pressure), and
the pipe elongation was calculated using the
small-deflection theory. A simple expression was
presented by the authors for calculating the longitudinal strain. Kennedy et al. (1977b) modeled
the pipeline as a flexible cable deforming into a
single curvature approaching asymptotically to
the undeformed portion. The pipe strain was calculated using the large-deflection theory considering a uniform passive earth pressure. Wang and
Yeh (1985) presented a refined analysis procedure
for calculating the elongation of buried pipelines
crossing either strike-slip or reverse strike-slip
faults using the large-deflection theory. One half
of the deformed pipe was assumed to be a curved
segment of constant curvature and a semi-infinite
segment on an elastic foundation. The model
involved the bending rigidity of the pipe, shear
force at the inflection point, and non-uniform
passive soil pressure, as well as the soil friction on
the pipe surface and tensile forces in the pipe
sections.
Soil liquefaction due to earthquakes of high
intensity and long duration is likely to occur
when the soil consists of loose uniform small-size
particles and becomes saturated. The assessment
of liquefaction hazards requires: (1) a qualitative
analysis that shows the susceptibility of different
soil conditions to liquefaction; (2) determination
of whether the anticipated strong earthquake
(which has 10% probability of occurrence) is
likely to induce liquefaction; and (3) determination of the length of the zone, the vertical and
horizontal soil displacements, as well as the
strength, density and viscosity of the soil. Only
very limited and approximate work has been done
on the analysis of pipeline response to soil liquefaction. An approximate estimation of pipe displacement was proposed by Kennedy et al.
(1977a) for pipelines buried within liquefiable soil.
Kennedy et al. (1977a), Wang and Yeh (1985),
Newmark and Hall (1975), Wang and ORourke
(1977) and Ariman and Muleski (1981) recommended that the following considerations account
for the design of buried pipelines crossing active
faults or running within liquefiable soil: (1) avoidance of placing the anchors, sharp bends or pipe

277

junction within a distance of at least 200 ft on


either side of the fault line to allow pipe movement in all directions; (2) shallow burial (35 ft)
with side slopes of 45 or less using cover and
back-fill from materials that have lower shear
strength; (3) use of thick-walled pipe without
abrupt change in the thickness to reduce the
strain concentration; (4) use of ductile steel, highquality welding and flexible joints to prevent
strain concentration and enable the pipe to tolerate large deformation; (5) allowing fault crossing
angle between 40 and 90 (the angle which places
the pipe in tension); (6) anchoring the pipe
against uplift or enhancing the pipe in concrete;
(7) use of smaller diameter pipes in parallel instead of one large diameter pipe; and (8) provide
fail safe systems at fault and liquefaction zones
(such as blow-off valves and reservoirs).

4. Seismic parametric behavior of buried pipelines


The seismic response of buried pipelines depends upon a number of factors. As a result,
parametric studies leading to the seismic behavior
of buried pipelines have been conducted by different investigators. Some of the important results of
the studies are reported below.
Shah and Chu (1974) observed that, if soil
structure interaction is ignored, then larger relative displacement at the bend results in higher
bending stress. Relative displacement between the
element and the surrounding soil increases with
an increase in particle velocity and decreases with
an increase in wave velocity. Relative displacement for a straight element forms an upper
bound for all cases. Therefore. this relative displacement should be used for finding bending
moments and shear forces in all cases for conservative estimate.
Datta et al. (1984) showed that response of the
pipe increases with the increase in the frequency
of P and SV waves with zero angle of incidence.
For other angles of incidence, the hoop stress
may decrease with the increase in frequency in
some cases. The pattern of stresses is depicted in
Fig. 5. The hoop stress also depends upon the
burial depth differently for different waves.

278

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

Fig. 5. Axial stresses for different angles of incidence.

Wong et al. (1985) reported that the tunnel (or


large diameter pipelines) behavior in rock and soft
soil are quite different. The hoop stress increases
with frequency initially and then decreases. If the
surrounding soil is soft, the maximum stress is
reached at a lower frequency. The dependence of
stress on the angle of incidence is found to be
quite different between P and SV waves, and
between soft and rock medium.
Hindy and Novak (1979) obtained the response
of buried pipelines crossing an interface between
two types of soils having different stiffnesses. The
highest stresses are found close to the boundary.
Takada (1977) studied the effect of pipe slippage
on the response behavior of buried pipelines. The
author recommended that flexible joints are necessary to absorb the sliding strain to assure earthquake resistance by harmonizing the pipeline to
the behavior of the ground.
The effect of pipe insulation, representing the
actual condition (where the pipe is usually surrounded by back-fill material), was studied by
Hindy and Novak (1980). They found that the
pipe insulation results in reduction of pipe
stresses. This result was confirmed by Datta et al.
(1984) using the shell model.
Datta and Mashaly (1986, 1988) observed that
about six or seven modes are sufficient for accurate analysis of both axial and lateral responses of
buried pipelines. The increase of radius and wall
thickness of the pipe reduces both axial and bending stresses. Structural damping may be ignored

compared to soil damping for calculating the


responses. The stresses near the pipe ends are
enhanced due to the pipe end conditions. The
degree of enhancement and the distance over
which the effect of end restraint prevails depend
upon the pipe end conditions. The axial and
bending stresses at the fixed ends are about 10.3
and 54 times more than those in the middle. For
pinned ends, they are greatly magnified over certain length near the ends as shown in Fig. 6
(Datta and Mashaly, 1988). The pipe stresses in
the middle are independent of end conditions and
remain constant over a substantial portion in the
middle as shown in Fig. 7 (Datta and Mashaly,

Fig. 6. Distribution of RMS bending stresses along the


pipeline.

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

Fig. 7. Longitudinal stresses in the buried pipeline.

1986). However, a minimum length of pipe is to


be considered in the analysis for each type of end
conditions in order to realize this result. The
embedment depth has little effect upon the
stresses. The pipe stresses are reduced by about
15% for shallow burial, and it becomes almost
constant for embedment depth more than 30
times the radius of the pipeline. Further, existence
of fluid inside the pipe does not significantly
influence the response unless sloshing effect becomes important.

5. Analysis of pipeline damage


Review of past earthquake damage records
(quantitative analysis) and typical modes of failure (qualitative analysis) enables designers to establish performance criteria to develop a
methodology that minimizes the probability of
system failure.
Observations from the 1971 San Fernando
earthquake showed that pipelines with rigid joints
failed more than those with flexible joints; most
damage was due to seismic shaking. The major
failure mechanisms were crushing, flexural failure
of pipe sections, pull-out and shear failure at the
joints. Failures due to buckling were particularly
dominant in the pipelines crossing a fault. They
occurred due to compressive forces generated by
the earthquake.
In the 1964 Alaska earthquake, it was observed
that buried pipelines were destroyed completely in
landslides. The differential settlement due to

279

earthquake shaking caused breakage of main and


service lines, while the direct seismic shaking
forces crushed and sheared the pipes, broke the
bells and pulled the mechanical joints apart.
Observed failures due to several earthquakes
since 1920 in Japan with magnitude greater than 7
showed that the damage by pipelines was mostly
caused by direct seismic shaking. In the 1964
Niigata earthquake, liquefaction had occurred
and caused most of the failures. The types of
failure due to these earthquakes were similar to
those observed during the San Fernando and
Alaska earthquakes.
After reviewing the data obtained from the
observation of damage due to past earthquakes,
Wang and ORourke (1977) concluded that: (1)
the damage occurred least in bed-rock, moderately in course-grained soils and most frequently
in fine-grained soil; (2) most damage occurred in
soils with response frequency between 3.5 and 4.5
Hz; (3) the damage increased for steel and cast
iron water pipes of small diameters, but for clay
sewer pipes the trend was reversed; (4) the bell
and spigot joints with lead or cement caulking
were vulnerable to earthquake effects and the
arc-welded pipes rarely failed at the girth welds;
and (5) asbestoscement pipes were generally
more susceptible to seismic damage. The welded
steel pipes behaved best.
The study by Katayama et al. (1975) on the
effect of ground motion intensity during the 1971
San Fernando earthquake showed that the number of failures per kilometer seemed to be linearly
proportional to the maximum ground acceleration
(loglog scale). This result was consistent with the
data of several earthquakes (Tokyo, Managua,
Fukui). Katayama et al. (1977) also studied the
effect of ground condition and complexity on the
damage of buried pipelines due to the 1923 Kanto
earthquake. It was observed that the damage index was greater for complex ground properties
and for very soft soil. In another study,
Shinozuka and Kawakami (1977) observed a
good correlation between the surface strain and
the damage index. An excellent state-of-art review
of the damage of buried pipelines under seismic
excitation was presented by Ariman and Muleski
(1981).

280

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

It was also observed that the effect of the


spatial correlation of ground motion has considerable effect on both stresses and displacement responses. The cross terms in soil damping and
stiffness have considerable effect on the frequencies and mode shapes of the pipeline but they do
not have much influence on stresses and displacements. This is because of the fact that the soil
pipe system is so stiff that the inertia effect
becomes negligible and quasi static response
governs.

6. Intersection stresses and network effect


Mashaly and Datta (1989a,b), ORourke et al.
(1982) and Singhal and Meng (1983) investigated
the intersection stresses in buried pipelines. It was
shown that the intersection stresses in the buried
pipelines depend upon a number of factors such
as the type of intersection. angle of intersection of
the pipes. the ratio of the diameters of the intersecting members and the angle of incidence of the
earthquake. Branches perpendicular to the direction of wave propagation have less stresses than
the other branches. Further, axial stresses are
much less than the bending stresses near the pipe
intersection. For a single straight segment of
pipeline, an opposite trend is observed. Generally.
the intersection stresses are much higher than the
stresses in the middle of branches: the amplification depends on the type of intersection, the angle
of incidence of earthquake and the ratio between
the diameters of the intersecting pipes. Typical
stresses at the pipe intersection are shown in Fig.
8. The variation of intersection stresses with the
angle of wave propagation for an L intersection is
shown in Fig. 9 (Mashaly and Datta, 1989a).
Shinozuka et al. (1979) have given conservative
values for the stresses at significant curvatures
and at pipe intersections, obtained by multiplying
the stresses in straight segments by a factor B
which ranges between 1 and 3. ORourke et al.
(1982) studied the effects of the propagation of
transverse seismic wave (S-wave) at junctions in a
buried pipeline network. They found that the
rotation in the straight pipe will be reduced if the
pipe junction exists. Shah and Chu (1974) devel-

Fig. 8. Intersection stresses.

oped a set of simplified expressions for calculating


the relative displacement between the soil and the
pipe, and shearing force and bending moment at
the pipe intersection. Singhal and Meng (1983)
analyzed pipe stresses as a quasi-static problem
using the idealization of beam on elastic foundation and using static loads acting at different
portions of the network. Results obtained by
Singhal and Meng (1983) and ORourke et al.
(1982) showed that very large bending stresses

Fig. 9. Variation of intersection stresses with angle of incidence (branch A2).

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

occur at the pipe intersections, and the network


effect extends to a distance ranging between pr0
and 40 r0 (where r0 is the radius of the pipeline).

7. Seismic risk analysis of buried pipelines


Seismic risk analysis of a buried pipeline system
is a methodology developed to minimize the probability of system breakdown and reduce the losses
from damage due to future earthquakes. Seismic
risk analysis requires the evaluation of the probability of pipeline system unserviceability under
different levels of earthquake intensity to which
this system is subjected during its lifetime. There
are many approaches to conduct the seismic risk
analysis, such as: (1) analysis for reliability of the
network system; (2) analysis for seismic hazard
losses; (3) vulnerability analysis; and (4) analysis
for system performance.
Taleb Agha (1977) employed the reliability and
network theories to conduct a seismic risk analysis of transportation lifelines. He developed
schemes to find the probability of partial network
failure due to a given set of ground shaking. Erel
et al. (1977) and Oppenheim (1979) developed a
relationship between the number of failures (N)
per kilometer of pipeline and peak ground acceleration or earthquake intensity. The damage conditions represented by N was translated into the
system performance. The system performance was
converted into losses and subsequently integrated
over the probability density function of earthquake occurrence to obtain the annual expected
loss. Oppenheim (1977) presented a direct numerical procedure using the Monte Carlo method to
simulate the failure of different lifeline links with
the magnitude of the earthquake and its expected
annual probability of occurrence. In the method,
the ground acceleration at each link due to each
incremental earthquake magnitude was calculated
and compared with the link resistance. For each
significant link failure, the performance of the
system in terms of losses was evaluated and total
expected annual hazard losses were obtained by
numerical summation.
Campbell et al. (1979) presented a methodology
for assessing the reliability of the lifeline in terms

281

of its effectiveness to meet certain minimum performance standards during and after an earthquake. These standards consider different risks to
the community. Mohammadi and Ang (1982)
conducted a seismic hazard analysis for lifeline
systems subjected to severe ground shaking. They
obtained the annual probability of failure of each
link of the system assuming uniform seismicity.
The probability of system failure was obtained by
modeling the lifeline using network approach.
Whitman and Hein (1977) presented a method for
seismic risk analysis of water pipeline systems of
shallow burial. In this analysis, a damage (unserviceability) probability matrix (DPM) was established for each segment using an appropriate
definition for damage state in terms of system
unserviceability. The elements of the DPM were
derived by combining two matrices. The first one
described the probability that each earthquake
intensity would produce different levels of failure,
and the other indicated the probabilities that each
level of soil site failure would produce different
damage states.
Despang and Shah (1982) proposed a method
for quantitative evaluation of seismic damage and
risk of lifelines and infrastructural systems. In this
analysis, the seismic damage was measured in
terms of global response, which includes many
measurements of physical damage and calibrations of the level of serviceability after the damaging event. Assuming that the response levels
respond as a Poisson stochastic process, the probability of n occurrences during a time T0 of system
global response larger than a certain value, G, is
obtained. The reliability is evaluated from the
conditional probability of the occurrence of the
state of G. Atkinson et al. (1982) proposed a
method for seismic risk analysis of buried pipelines assuming that the seismic waves are fully
correlated and travelling along the pipe axis with
surface Rayleigh wave. The axial stress was obtained by neglecting the soilpipe interaction.
Shinozuka et al. (1979) developed a methodology for the seismic risk analysis of water transmission systems. The damage states were defined by
some levels of axial pipe strain which were calculated considering ground shaking, fault movement
and soil liquefaction. The probability of different

282

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

damage states was calculated for every combination of soil condition and earthquake intensity
considering the pipe strain as a Gaussian random
variable. These probabilities were tabulated in
DPMs as presented by Whitman and Hein (1977).
Mashaly and Datta (1989c) described a procedure
for seismic risk analysis of the component segment of the general network system of buried
pipelines. The concept of DPM was used to obtain an estimation of the annual probability of
occurrence of different damage states. The response of the pipeline was obtained by using the
method of random vibration analysis.

3. Evaluation of stresses produced in large diameter pipes due to random ground motion
caucing buckling and fracture failures.
4. Stresses induced in the pipeline network system produced by fault movements and in
pipelines passing through very soft and liquefiable soil.
5. Qualitative and quantitative evaluation of
damages in buried pipelines based on the information available from more recent earthquakes, and consequent seismic risk analysis
of pipeline network systems.

References
8. Conclusions
A state-of-the-art review of the seismic behavior
of buried pipelines is presented. It also includes
information on the estimation of pipe damage
caused by seismic forces and the consequent seismic risk analysis. The review shows that considerable research has been carried out on the
determination of pipe response to traveling (deterministic) seismic waves; in particular, many simplified formulae have been developed to obtain
seismic stresses produced in the pipeline. Comparatively much less attention has been paid to the
evaluation of pipe stresses for random earthquake
inputs and for network systems crossing active
faults. Specifically, the stresses at the intersection
of network systems under random seismic conditions, which are important for seismic risk analysis, have not received much attention. It is felt
that more investigations are necessary on the following subjects in order to formulate rational
design procedures and to obtain realistic assessment of seismic risks involved in the designs:
1. Network effect on the stresses in the pipeline
with special reference to the evaluation of the
stresses at pipeline intersections produced by
random ground motion.
2. Experimental and field data procurement and
analysis to establish realistic behavior of soil
pipe interactive forces, behavior of pipe joints,
and actual strain and deformation patterns in
the pipe produced by ground motion.

Ariman, T., Muleski, G.E., 1981. A review of the response of


buried pipelines under seismic excitations. Earthquake
Eng. Struct. Dyn. 9, 133 151.
Atkinson, G.M., Davenport, A.G., Novak, M., 1982. Seismic
risk to pipeline with application to northern Canada. Can.
J. Civil Eng. 9, 248 264.
Campbell, K.W., Enguchi, R.T., Duke, C.M., 1979. Reliability
in lifeline earthquake engineering. J. Tech. Counc. ASCE
105, 259 269.
Clough, R.W., Penzien, J., 1975. Dynamics of Structures.
McGraw-Hill, New York.
Datta, T.K.., Mashaly, E.A., 1986. Pipeline response to random ground motion by discrete model. Earthquake Eng.
Struct. Dyn. 14, 559 572.
Datta, T.K., Mashaly, E.A., 1988. Seismic response of buried
submarine pipelines. Trans. ASME J. Energy Resour.
Technol. December, 208 218.
Datta, S.K., Shah, A.H., El-Akily, N., 1981. Dynamic behavior of buried pipeline in seismic environment. Trans.
ASME J. Appl. Mech. 49, 141 148.
Datta, S.K., Shah, A.H., Wong, K.C., 1984. Dynamic stresses
and displacements in buried pipe. J. Eng. Mech. ASCE
110, 1451 1465.
Despang, H.M., Shah, H.C., 1982. Seismic risk analysis of
infrastructural systems. J. Techn. Counc. ASCE 108, 10
23.
Erel, B., Patelunas, G.M., Niece, J.E., Oppenheim, I.J., 1977.
Measuring the earthquake performance of urban water
systems. The current state of knowledge of lifeline. Earthquake Eng. J. Struct. Eng. ASCE 103, 183 198.
Hall, W.J., Newmark, N.M., 1977. Seismic design criteria for
pipelines and facilities. The current state of knowledge of
lifeline: earthquake engineering. J. Struct. Eng. ASCE 103,
18 34.
Hindy, A., Novak, M., 1979. Earthquake response of underground pipelines. Earthquake Eng. Struct. Dyn. 7, 451
476.

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284


Hindy, A., Novak, M., 1980. Pipeline response to random
ground motion. J. Eng. Mech. ASCE 106, 339 360.
Katayama, T., Kubo, K., Sato, N. Earthquake damage to
water and gas distribution systems. Proc. U.S. Natl.
Conf. Earthquake Eng., 1975, pp. 396405.
Katayama, T., Kubo, K., Sato, N., 1977. Quantitative analysis of seismic damage to buried utility pipelines. In:
Proceedings Sixth World Conference Earthquake Engineering. Institute Association Earthquake Engineering,
New Delhi, pp. 33693375.
Kennedy, R.P., Darrow, A.C.., Short, S.A., 1977a. General
considerations for seismic design of oil pipelines systems.
The current state of knowledge of lifeline: earthquake
engineering. J. Struct. Eng. ASCE 103, 217.
Kennedy, R.P., Chow, A.W., Williamson, R.A., 1977b.
Fault movement effects on buried pipelines. J. Transport.
Eng. ASCE 103, 617633.
Kubo, K., Katyama, Ohasi, M., 1979. Lifeline earthquake
engineering in Japan. J. Technol. Counc. ASCE 105,
221 238.
Kuesel, T.R., 1969. Earthquake design excitation for subways. J. Struct. Eng. ASCE 95, 12131231.
Lee, V.W., Trifunac, M.D., 1979. Response of tunnels to
incident SH-waves. J. Eng. Mech. ASCE 105 (4), 643
658.
Luco, J.E., Barnes, F.C.P., 1994. Seismic response of cylindrical shell embedded in layered visco elastic half space.
Earthquake Eng. Struct. Dyn. 23, 553580.
Mashaly, E.A., Datta, T.K., 1989a. Seismic stress at the
intersection of buried pipelines. J. Pipelines 7, 281 299.
Mashaly, E.A., Datta, T.K., 1989b. Seismic behavior of
buried pipelines. J. Pipelines 7, 215234.
Mashaly, E.A., Datta, T.K., 1989c. Seismic risk analysis of
buried pipelines. J. Transport. Eng. ASCE 115 (3), 232
251.
Masso, A.G., Attalla, I., 1984. Finite element verses simplified method in the seismic analysis of underground
structures. Earthquake Eng. Struct. Dyn. 12, 347 367.
Mohammadi, J., Ang, A.H.S., 1982. Seismic hazard analysis
of lifelines. J. Struct. Eng. ASCE 108, 12321249.
Muleski, G.E., Ariman, T., Aumen, C.P., 1979a. A shell
model of buried pipe in a seismic environment. Trans.
ASME J. Pressure Vessel Technol. 101, 1450.
Muleski, G.E., Ariman, T., Aumen, C.P., 1979b. A shell
model for buried pipes in earthquakes. J. Soil Dyn.
Earthquake Eng. 4, 4351.
Nelson, I., Weidlinger, P., 1979. Dynamic seismic analysis of
long segmented lifelines. Trans. ASME J. Pressure Vessel
Technol. 101, 1020.
Newmark, N.M., Rosenblueth, E., 1971. Fundamentals of
Earthquake Engineering. Prentice-Hall, Englewood Cliffs.
Newmark, N.M., Hall, W.J. Pipeline design to resist large
fault displacement. Proc. US Natl. Conf. Earthquake
Eng. Res. Inst., 1975, pp. 24162425.
Novak, M., Hindy, A., 1980. Seismic analysis of underground tubular structures. Proc. Seventh World Conf.
Earthquake Eng. 8, 287294.

283

OLeary, P.M., Datta, S.K., 1985a. Dynamics of buried


pipelines. J. Soil Dyn. Earthquake Eng. 4, 151 159.
OLeary, P.M., Datta, S.K., 1985b. Dynamic response of
buried pipeline at low frequencies. Trans. ASME J. Pressure Vessel Technol. 107, 44 50.
Oppenheim, I.J., 1977. Vulnerability of transportation and
water systems to seismic hazard. methodology for hazard
cost evaluation. The current state of knowledge of lifeline. Earthquake Eng. J. Struct. Eng. ASCE 103, 394
409.
Oppenheim, I.J., 1979. Simulation of water system seismic
risk. J. Technol. Counc. ASCE 105, 327 336.
ORourke, M.J., Pikul, R.R., Wang, L.R.L., 1982. Transverse seismic waves at pipeline junctions. J. Tech. Counc.
ASCE 108, 173 177.
Shah, H.H., Chu, S.C., 1974. Seismic analysis of underground structural elements. J. Power Div. ASCE 100,
53 62.
Shinozuka, M., Kawakami, H., 1977. Underground pipe
damages and ground characteristics. The current state of
knowledge of lifeline. Earthquake Eng. J. Struct. Eng.
ASCE 103, 293 307.
Shinozuka, M., Takada, S., Ishikawa, H., 1979. Some aspect
of seismic risk analysis of underground lifeline systems.
Trans. ASME J. Pressure Vessel Technol. 101, 31 43.
Singhal, A.C., Meng, C.L., 1983. Junction stresses in buried
jointed pipelines. J. Transport. Eng. ASCE 109, 450461.
Singhal, A.C, Zuroff, M.S., 1990. Analysis of underground
and underwater space frame with stiff joints. J. Comput.
Struct. 35 (1), 227 237.
Takada, S., 1977. Earthquake resistance design of underground pipelines. In: Proceedings Sixth World Conference
on
Earthquake
Engineering.
International
Association Earthquake Engineering, New Delhi, pp.
3376 3381.
Takada, S., Tanabe, K., 1987. Three dimensional seismic
response analysis of buried continuous or jointed pipelines. J. Pressure Vessel Technol. ASME 109, 80 87.
Taleb Agha, G., 1977. Seismic risk analysis of lifeline networks. Bull. Seismol. Soc. Am. 67, 1625 1645.
Wang, L.R.L., ORourke, M.J., 1977. State of the art of
buried pipeline earthquake engineering, the current state
of knowledge of lifeline: earthquake engineering. J.
Struct. Eng. ASCE 103, 252 265.
Wang, L.R.L., Yeh, Y.H., 1985. A refined seismic analysis
and design of buried pipelines for fault movement.
Earthquake Eng. Struct. Dyn. 13, 75 96.
Wang, L.R.L., ORourke, M.J., Pikul, R.R., 1979. Seismic
response behavior of buried pipelines. Trans. ASME J.
Pressure Vessel Technol. 101, 21 30.
Wang, L.R.L., Pikul, R.R., ORourke, M.J., 1982. Imposed
ground strain and buried pipelines. J. Technol. Counc.
ASCE 108, 259 263.
Whitman, R.V., Hein, K.H., 1977. Damage probability for a
water distribution system, the current state of knowledge
of lifeline. Earthquake Eng. J. Struct. Eng. ASCE 103,
410 423.

284

T.K. Datta / Nuclear Engineering and Design 192 (1999) 271284

Wong, F.S., Weidlinger, P., 1983. Design of underground


protective structures. J. Struct. Eng. ASCE 109, 1972
1979.
Wong, K.C., Shah, A.H., Datta, S.K., 1985. Dynamic stresses
and displacements in a buried tunnel. J. Eng. Mech. ASCE
111 (2), 218234.

Wong, K.C., Shah, A.H., Datta, S.K., 1986. Three dimensional motion of buried pipeline. J. Eng. Mech. ASCE 112,
1319 1348.
Yuan, H.R., Walker, R.E., 1970. The investigation of simple
soil structure interaction model. In: Dynamic Waves in
Civil Engineering. Wiley, New York, pp. 241 266.

You might also like